How I work
Demand first, then one niche, start cheap, test for three months, kill side projects around €1,000.
01
The moment
Two weeks. That is how long it took from my approved Upwork profile to the first client. Not because I was the best at everything, but because I had a clear niche: tracking and data. What the job paid, I will add.
Those two weeks turned into rules I have worked by since. If you are thinking about how to get out of your job right now, you probably face the same questions I faced then.
The year of that first job goes here once added.
02
What went wrong
I have broken these rules myself, which is why they are here. The latest example is the learning center with daily courses, whose figures are under Repeat. I built it without checking demand first. That is exactly the first rule, and I did not keep it.
The mistake behind it: with AI you can start more in a week than you used to in a year, and the learning center was built just that fast. Whether anyone wanted it, I had not asked beforehand.
03
What I know now
What I know now is five rules, in the order in which I apply them. The first decides whether I start at all. The last decides whether I stop.
Research demand first. Before I build anything, I check whether someone searches for it and pays for it. That is not gut feeling but research. What do people search for? What do they already pay for today, maybe to someone else? If I find no answer, I do not build.
Build the offer around one niche. I like to say: rather a brain surgeon than a general practitioner. Whoever solves one thing for one group gets found and recommended. Whoever does everything for everyone gets compared and pushed down on price. With AI a niche can be smaller today than it used to be, because one person can serve a very specific group very well when the routine barely costs time.
Start cheap and overdeliver. The first clients are proof, not profit. They show you whether the offer holds, and they recommend you. You can raise the price once you have that. The other way round almost never works.
Test one service for three months. Shorter tells you nothing, because the first weeks always look too good or too bad. Longer costs time a better offer could have had. Three months are long enough to see a pattern and short enough to stop in time.
Kill side projects that earn around €1,000. They feel like success, because money comes in, just enough that you never switch them off. And they keep you from the actual business, because they cost just as much attention as something that could bring ten times as much.
These five rules belong together. Demand tells you whether the market exists. The niche tells you who your client is in it. The cheap start brings the first proof. The three months tell you whether it carries. And ending the small side projects makes sure you have the time to actually do all of it. Leave one out and the others wobble. Four of the five rules are a no, which is what makes them uncomfortable.
04
Open question
What is open is whether three months are still the right span in a world that changes this fast. Maybe today you have to decide faster. Maybe it is the other way round, because everyone switches so fast that stamina becomes rare and therefore valuable.
05
Changelog
First version. Nothing here is rewritten silently: every later change gets its own entry in this changelog.